Mechanical Engineer & Python Expert - Freelance AI Trainer

Role overview

This is a project-based opportunity with Mindrift, where subject-matter experts help evaluate and improve AI systems for major technology companies. It is not a permanent position and participation is organized around individual projects.

What you will do

You will create computational engineering challenges that push a frontier AI model to solve domain-specific problems. Each task must have a code-checkable answer and must rely on a specialized engineering tool such as Cantera, CoolProp, CalculiX, OpenFAST, or another comparable package. Generic numerical libraries alone are not sufficient. Every problem is executed in a locked Linux environment with the required tool already installed, and a programmatic grader evaluates the model's response.

• Choose a core tool and build a task that depends on its solver, simulation engine, or specialized domain model.
• Create a Python-based reference answer and provide any needed input files, geometry, or mechanism definitions.
• Set the target result and define an appropriate tolerance so the answer can be judged correctly.
• Run the task against the model in parallel test batches, then refine it until success rates stay in the 10–30% range.
• Submit the polished task for review by a senior specialist in your field, who will assess quality and provide feedback.

Task calibration

Tuning the difficulty requires iteration and patience. You may need to adjust thermodynamic cycles, material models, and boundary conditions while observing where the model succeeds, where solvers converge, and where simulations fail. The work also helps deepen your practical knowledge of the chosen engineering tool and provides insight into how advanced models approach thermal, structural, and fluid mechanics problems.

Candidate profile

This role suits engineers with Python experience who are open to part-time, non-permanent project work. Strong written English is required, and you should be prepared to work independently when learning new tools.

Application note

Please send your CV in English and include your English proficiency level.

How the workflow operates

The process is: apply, complete qualification steps, join a project, complete tasks, and receive payment.

Workload and compensation

During active project periods, the expected workload is around 10 to 20 hours per week. This figure is only an estimate and is not guaranteed. Pay can reach the equivalent of $50 per hour depending on your experience, pace, and the project scope. Compensation differs from one project to another based on complexity and required expertise.

Additional information

Priority is given to people who can build problems that truly require a specialized solver. Experience with the listed engineering tools is helpful, but not mandatory if you can learn quickly and start contributing independently.

In order to be considered, applicants should have at least a degree in Mechanical Engineering or a closely related discipline, along with 2+ years of research, applied, or teaching experience.

The opportunity is intended for contributors who are open to project-based collaboration rather than permanent employment.